论文部分内容阅读
本文给出一种既能有效衰减地震噪音又可保护地层及构造的不连续性的新方法。构造约束保边平滑技术需要已知反射局部方位和边界信息,通常这些信息由全频率地震资料估算获得,但在资料信噪比很低的情况下,噪音往往会降低估算的可靠度。对于信噪比极低的地震资料,其主频成分相对非主频成分信噪比高,所以由主频资料获取的方位和边界信息比由其它频率成分获取的更可靠。方位和边界信息通常用倾角和相干值差异来描述。由于不同频率所引起的倾角和相干值差异的变化均比地震记录的变化缓慢,所以由主频资料获取的倾角及边界信息能够近似代表所有频率成分的倾角及边界信息。Ricker子波广泛用于地震勘探,Marr小波与Ricker子波在时间和频率域均具有相同的形态,所以选用Marr小波变换将地震数据按照倍频程分为几个分频体。扫描主频分频体,用不等权二次曲面拟合并求解极大值来获取视倾角,通过比较9个滑动窗口的相干值来确定反射边界。将这些信息用构造约束保边平滑技术可选择性地(selectively)对主频、低频、高频分频体做平滑处理,最后将平滑后的各频段地震记录合成为滤波去噪后的地震记录。理论模型和实际资料处理效果表明该方法能有效压制噪音,保护边界,保护同相轴的连续性,且灵活地保留地震记录中的有用信息。
This paper presents a new method that can not only effectively attenuate the seismic noise but also protect the continuity of the formation and structure. Construction Constraints Paul-preserving smoothing techniques require known local reflection and boundary information, which are usually estimated from full-frequency seismic data. However, at very low signal-to-noise ratios, noise tends to reduce the reliability of the estimation. For seismic data with extremely low signal-to-noise ratio, the dominant frequency component has a higher signal-to-noise ratio than the non-dominant frequency component, so the orientation and boundary information obtained from the dominant frequency data is more reliable than those obtained from other frequency components. Orientation and boundary information is usually described by differences in dip and coherence. Since the variation of dip and coherence caused by different frequencies is slower than the change of seismogram, the dip and boundary information obtained from the dominant frequency data can approximately represent the inclination and boundary information of all frequency components. The Ricker wavelet is widely used in seismic exploration. The Marr wavelet and Ricker wavelet have the same shape in both time and frequency domain. Therefore, Marr wavelet transform is used to divide the seismic data into several sub-bands according to the octave band. Scan the main frequency divider, obtain the apparent dip by fitting the quadratic surface with unequal power and solving for the maximum value, and determine the reflection boundary by comparing the coherence values of the 9 sliding windows. This information can be selectively smoothed by frequency-domain, low-frequency and high-frequency crossover using the constructed constrained edge-preserving smoothing technique. Finally, the smoothed seismic records are synthesized into the filtered and de-noised seismic records . The theoretical model and actual data processing results show that this method can effectively suppress the noise, protect the boundary, protect the continuity of the events, and flexibly retain the useful information in the seismic record.